Implicit Gaze+Slide: Discrete and low-effort typing for MR using gaze-based prediction and finger motion refinement

Current commercial Mixed Reality (MR) headsets allow users to run applications in a variety of use contexts. When text entry is required, common MR techniques rely on visible hand gestures in mid-air to select keys on a virtual keyboard that can cause fatigue and draw attention from bystanders. Even techniques that use eye-gaze to minimize hand motions depend on visible hand gestures in mid-air for selection, as well as adding unintentional effort when users try to control their gaze direction. We present a novel interaction technique, Implicit Gaze+Slide aimed at easy and private text entry. First, it measures eye gaze without rendering the explicit gaze location to the user, enabling a fast and low-effort selection of a keyboard area. Then a one-handed finger slide over a phone surface is used for fine selection. In contrast to previous works, the hand interaction is not viewed by the user or the glasses, and can be hidden in a pocket from standers. In a controlled user study comparing Implicit Gaze+Slide with an explicit-gaze baseline, Implicit Gaze+Slide achieved a comparable entry rate (11.51 wpm), a lower raw character error rate (5.49%), and lower perceived physical and mental demand. In a separate privacy-perception survey, Implicit Gaze+Slide was rated as substantially more discreet than representative mid-air alternatives.